Metacognitive Monitoring and Control in Autistic Eyewitnesses

Background:

Providing eyewitness testimony involves effectively monitoring one’s memory in order to provide the most detailed, yet accurate account possible; reporting details that one is sure of and withholding details that may be inaccurate. According to Koriat and Goldsmith’s (1996) monitoring and control framework, an individual will provide fine-grained detail (e.g., navy blue) when their confidence of accuracy exceeds a criterion threshold; if not, they retrieve a less detailed, coarser-grained response (e.g., dark) (Ackerman & Goldsmith, 2008). Autistic individuals experience difficulties both in retrieving specific episodic memories (Bowler et al., 2000) and in monitoring the accuracy of their output (e.g., Grainger, Williams, & Lind, 2016). They also show a bias towards local, detail-focused processing (Mottron et al., 2006).

Objectives:

1. Do autistic and typically developing (TD) witnesses differ in monitoring the accuracy of their output?
2. Do autistic and TD witnesses differ in strategically controlling their output, weighting accuracy against informativeness?
3. Do both autistic and TD witnesses maximise accuracy over informativeness when explicitly instructed to do so?

Methods:

Based on an a-priori sample size calculation, 34 autistic adults and 34 age- and IQ-matched TD adults viewed a short video clip of a mock bank robbery before answering questions about it. In Phase 1, participants freely generated the granularity of their responses (i.e., fine- or coarse-grained). In Phase 2, participants answered the same questions again but this time they were asked to provide both a fine-and a coarse-grained answer to each question. Finally, in Phase 3, participants were shown their answers from Phase 2, and were instructed to select one of these as their final answer, and they were asked to maximise accuracy over informativeness by only reporting fine-grained details if they were absolutely confident they were correct. Confidence ratings were taken for each response in all phases.

Results:

Data collection is in the final stages but preliminary results from the existing data indicate that both groups showed a strong free reporting preference for fine-grained details in Phase 1 (ASD 87.5% fine-grained; TD 95.6%) and did so with similar rates of accuracy: ASD M accuracy =0.42, SD=0.17; TD M accuracy =0.43, SD=0.10 (p=.768). While autistic and TD witnesses showed similar confidence-accuracy correlations for reported fine-grained details (ASD M Gamma=.45, TD M Gamma=.46), there was some preliminary evidence that TD witnesses were better at judging the accuracy of coarse-grained information (M Gamma=.63) than autistic witnesses (M Gamma=.49), suggesting potentially subtle differences in the monitoring and regulation of output. Nevertheless, all witnesses were significantly more accurate in Phase 3 when explicitly instructed to maximise accuracy over informativeness, F(1,30)=34.32, p<.001, ηp²=.53, and there was no Group x Phase interaction, F(1,30)=0.98, p=.331, ηp²=.03, indicating that autistic and TD witnesses benefited similarly from this instruction. The improvement in accuracy was driven by an increase in coarse-grained reporting in Phase 3, F(1,30)=317.14, p<.001, ηp²=.91.

Conclusions:

Findings have implications for the instructions that witnesses receive, with preliminary findings indicating that autistic witnesses can strategically regulate their output to maximise accuracy when explicitly instructedto do so.